Open AccessAn Auxiliary Identification System for Destructive Pests
Author(s) -
Weijian Mai,
Fengjie Wu,
Jiangtao Chen,
Yisheng Tang,
Qingkun Liu
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2010/1/012095
Subject(s) - identification (biology) , convolutional neural network , benchmark (surveying) , cosine similarity , latent dirichlet allocation , similarity (geometry) , computer science , artificial intelligence , government (linguistics) , resource (disambiguation) , machine learning , pattern recognition (psychology) , image (mathematics) , data mining , topic model , geography , cartography , computer network , linguistics , philosophy , botany , biology
This work takes this Asian Hornet’s invasion of Washington in 2020 as an example to build an auxiliary identification system for Destructive Pests to help the government’s rescue work. To help the government to prioritize the resource to the most possible reports and improve the investigation efficiency, we built a binary classification model to classify the report that contains images and text. For image classification, we design a VGG16-based convolutional neural network pretraining with ImageNet, which achieved superior performance with a high F1 score (95.08%) on our dataset. For text classification, we use the Latent Dirichlet Allocation (LDA) model to cluster words into different themes and adopt cosine similarity to calculate the similarity between the new report and the benchmark positive/negative sets. The results demonstrated that our approach can identify target pests efficiently and help the government to prioritize the resource to the most possible reports, improving the investigation efficiency.